Developed a multipurpose AI chatbot using Flask and Chainlit with a popup chat UI for seamless interaction.

Integrated LangChain-based RAG pipeline with FAISS vector store, Groq LLMs, and PDF document ingestion via HuggingFace embeddings.

Built for domain-specific Q&A, supporting healthcare and technical documents with fast semantic search and real-time responses. 

Developed a real-time Facial Emotion Detector using FastAPI and MediaPipe for accurate face tracking.

Integrated multi-provider AI inference with Hugging Face models, Groq LLMs, and OpenRouter APIs to support 7-class emotion recognition.

Deployed on Render with Docker for production-ready scalability, featuring live webcam analysis, image uploads, and interactive UI with Tailwind CSS. 

It's an AI-powered assistant built using Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG).

It analyzes PDF service manuals of biomedical machines to provide precise repair and troubleshooting support.

Designed to help clinical engineers quickly resolve device issues with step-by-step, context-based guidance.

Engineered personalized book recommendations using collaborative and content-based filtering. Using NLP techniques and tools such as Python, NLTK, and SpaCy to extract insights and similarities between the books.

The analysis was conducted efficiently, saving time and eliminating the need for manual data processing. The outcome was the ability to recommend similar books based on similar books and genres.

Extend the parser to handle various resume formats (e.g., PDF, DOCX, TXT) to accommodate a wider range of submissions. 

Implemented an NLP model to extract and structure resume content, improving HR efficiency. Customized entity recognition and skill matching for precise candidate selection. 

Tailored entity recognition and skill matching, ensuring the parser selects candidates with utmost accuracy for a refined recruitment process. 

Applied advanced algorithms to identify global socio-economic patterns.

Uncovered actionable insights for targeted policy interventions by determining distinct clusters within world development data

Applied advanced feature selection techniques to enhance cluster accuracy and interpretability in identifying global socio-economic patterns. 

Embark on a data-driven journey with this Nifty50 web scraping repository, where Python's Requests and BeautifulSoup libraries join forces. 

Streamlining HTTP requests and HTML parsing, this dynamic duo enables developers to effortlessly retrieve and extract precise data from Nifty50 websites, empowering efficient and effective web scraping endeavors.